The rising trend of Robots as a Service (RaaS)



This article is devoted to the historical overview of the Robot-as-a-Service concept. Several major scientific publications on the development of Robot-as-a-Service systems based on a service-oriented paradigm are considered. Much attention is paid to the analysis of a centralized approach in the development using cloud computing services and the search for the limitations of this approach. As a result, general conclusions on the reviewed publications are given, as well as the authors' own vision of Robot-as-a-Service systems based on the concept of robot economics.To get more news about Robotics as a Service, you can visit glprobotics.com official website.

1. Introduction
The automation and robotization trend that began in the second half of the twentieth century is now moving into a qualitatively new stage. Due to the widespread adoption of the Internet, mobile devices, sensors, and video cameras, as well as deep learning methods, almost every corner of a large city becomes digitized (Lyons et al., 2018). The proliferation of unmanned aerial vehicles (Ortiz et al., 2018) and cheap and high-performance satellite Internet channels (Sayin et al., 2019) promises even more extensive digitalization, now on a global scale. Engineering also does not stand still, and in recent years, developers have achieved outstanding results that open up new opportunities for using robots in everyday life (Nelson et al., 2018).

These changes cannot but affect the most important sphere of human welfare—the economy. The development of digital technologies has led to “uberization” of the relationships between the client and the service providers (David et al., 2016). The as-a-service business model has completely dominated the software industry in the past several years. It has grown to other sectors now, where continuous recurring revenue has replaced the one-time purchases.
2. Robot-as-a-Service Concept
As already mentioned, the concept of Robot-as-a-Service appeared as part of a service-oriented paradigm, continuing the list of “as-a-Service” concepts (Platform-as-a-Service, Software-as-a-Service) (Blokdyk, 2018). In short, the hallmark of these concepts is the refusal to purchase hardware or software directly. Instead, it is proposed to receive all the necessary services by subscribing to them. RaaS involves the rental of robotic devices (often expensive) with the ability to deploy your applications and services.

Typically, the RaaS platform contains: basic services that describe the functionality of the robot; the ability to add and select user services; standardized communication protocol (e.g., Web Services Description Language, Simple Object Access Protocol, HTTP); integration with a computing environment and a database for performing complex calculations and storing information (Blokdyk, 2018). The last item in the first and most popular implementation of RaaS is represented by cloud computing services (from Amazon, Google, Microsoft, and other enterprizes). An exemplary description of the RaaS system's operation through the cloud is shown in Figure 1.
3. Historical Overview of RaaS Publications
This overview was done by searching with the keyword for “robot-as-a-service” in scientific databases. The purpose of this review is to give an idea of how progress in this area. In addition to a general overview of the technology, we look primarily at how authors integrate their work into the economy.

3.1. Robot as a Service in Cloud Computing, 2010
The most cited publication on the RaaS concept and one of the first publications to use this term (Chen et al., 2010). For a long time, the authors were engaged in service-oriented architecture and cloud computing together with Intel and Microsoft, and at that time, it was logical to resort to the dominant SOA approach with respect to RaaS.

In their understanding, RaaS should have all the functions presented in SOA: service provider, service broker, service client. That is, for each device there is a repository with downloadable services, all services are available for the client in a shared directory, and a new application can be created based on existing ones.

As a result, the RaaS unit is a robot for which all the described functions are implemented, and access to them is in the cloud. The cloud is the provider, and the broker is an interactive shell that allows us to view and select any of the services. Interaction between units occurs through a standard interface (WiFi, Bluetooth), and between services through a Web Services Description Language. The authors devote much time to their RaaS prototype on Arduino, Intel, and Lego NXT boards and show how such an architecture can be ported to various platforms (Java on Linux, C#, and Visual Programming Language on Windows).

3.2. Internet of Intelligent Things and Robot as a Service, 2013
This article is from the same authors (Chen and Hu, 2013) and it presents a more extended concept of RaaS focused on the Internet of Things (IoT). The authors emphasize a critical feature of RaaS: they note the cyclical nature of the changes in centralized and decentralized paradigms for constructing such systems. Noting the disadvantages of decentralization, the authors nevertheless explain that decentralized elements are needed to increase the centralized system's reliability. Among the typical narrow points for RaaS systems, they present scalability, service orientation, security, adaptability to changes, and fault tolerance.
4. Discussion
As we can see, when it comes to building Robot as-a-Service systems, the SOA concept based on cloud computing is practically the only one that is being researched and implemented. The difference is only in the details of the RaaS implementation, such as various communication protocols.

The overall architecture remains highly centralized, despite attempts to introduce decentralized elements. This is, in our opinion, a critical place for the implementation of RaaS systems. Studies show (Cummings, 2015) that decentralization reduces the computational burden on the agent management system and allows us to connect more devices than a centralized one. Moreover, large cloud services are highly centralized structures, which can adversely affect the security of RaaS systems. In the event of a center failure or hacking, the multi-agent system of robots will simply cease its activity at best. The worst-case scenario involves taking control of robots and devices and subsequent misconduct.